Neuro-Symbolic AI Aims to Make Legal Reasoning Trustworthy
A new research proposal outlines a neuro-symbolic approach to legal AI that combines large language models with formal verification. The goal is to address the problem of LLMs drawing assumption-laden conclusions beyond what source texts support, which poses serious risks in high-stakes legal work. The system aims to reduce manual verification burdens while maintaining accountability.
Key facts
- Proposal combines large language models with formal verification
- Addresses LLMs drawing unsupported inferences in legal contexts
- Aims to reduce manual verification burden
- Targets high-stakes legal practice requiring rigor
- Published on arXiv under Computer Science > Artificial Intelligence
Entities
Institutions
- arXiv